Annualised return
The compound growth rate implied by where your portfolio started and finished, expressed per year. It makes periods of different lengths comparable, and hides everything that happened in between.
On this page
Annualised return converts a total return over any period into a yearly rate.
It exists because a raw return is uncomparable across time. Gaining 40% is remarkable over eight months and unremarkable over eight years, and the number 40% cannot tell you which you are looking at. Annualising divides that out.
How it is calculated
annualised return = (final value / starting value) ^ (252 / number of observations) − 1
Gylder takes the first and last value in your selected window and works out the compound rate that would connect them, scaled to a year of 252 trading days.
Compound, not average. A portfolio that gains 50% then loses 50% has not broken even, and annualising by compounding reflects that correctly where averaging would not.
How to read it
It is the constant yearly rate that would have produced the same result. Nothing more.
That "constant" is doing real work, and it is the measure's main limitation. Two portfolios with identical annualised returns can have had completely different journeys: one drifting up steadily, one collapsing and recovering. The figure is blind to the path, which is precisely why volatility and maximum drawdown sit next to it.
Read it with them, never on its own.
Where it is used
Annualised return is the numerator of the Calmar ratio, which divides it by your worst fall.
It is also the figure to use when comparing periods of different lengths, which is the one job total return cannot do.
Two cautions
Short windows exaggerate. Annualising a strong month projects that month across a year, which produces impressive figures with almost no evidence behind them. A 4% month annualises to roughly 60%, and nobody should read that as a yearly expectation.
It uses two points. The first and last values decide it. If your window happens to start at a low point or end at a high one, the figure inherits that framing, which is another reason changing the date range moves it so much.
Not a forecast
An annualised return describes what happened. It is not a prediction, and Gylder does not present it as one or project it forward.
For a return figure that accounts for when you actually put money in rather than only where the portfolio started and ended, see time-weighted vs money-weighted return.
Was this helpful?